The Effect of Different Ability Distribution on Item Parameters and Test Information Function

Abstract

This study aimed at verifying the effect of the ability (θ)
distribution on item parameters and Test Information Function (TIF).
To achieve the aim of the study, four ability distributions' were
generated according to three parameter logistic model using
WINGEN program on 60 dichotomous items. The responses of the
examinees were analyzed by BILOG– MG 3 programs for each ability
distributions according to the three parameter logistic model. The
results of the repeated measurement ANOVA revealed that there were
statistical significant differences between difficulty means and
discrimination means of the scaled items from ability distribution. In
spite of the variance of item parameters, the results revealed that the
correlation between the scaled items with different ability were
statistically significant as indicators of item parameters stability
among different distributions of ability. The results also revealed that
the Test Information Function varied with the variance of the ability
distribution, but they all gave the highest values of the items function
within the range of ability (0.5-0.63), where the measurement’s
standard errors were the least .